Page 196 - Biaxial Multiaxial Fatigue and Fracture

180 f? BONACUSE AND S. KALLURI

For many of these model categories, the experimental results discussed in this paper present
various difficulties, both in interpretation and implementation. Both the LDR and the DCA
have been shown to be less than ideal predictors of the observed interaction behavior. As
Haynes 188 has been shown not to have an endurance limit, many of the life curve
modification models that incorporate this concept would tend to be inaccurate in many regimes.
Crack growth approaches may not prove useful in describing this data set in light of the
observed damage accumulation behavior in the mixed loading experiments. The final failure
crack direction is only a function of the final loading direction (shear cracking for the
axidtorsion experiments and maximum normal stress cracking for the torsiodaxial
experiments). The axial/torsion and torsiodaxial experiments seem to closely follow the LDR
(with the exception of the highest applied life fraction in the axidtorsion experiments). The
mixed loading experiments performed in this study seem to indicate that for Haynes 188 at
538"C, there is no unusual interaction under the conditions imposed. Models that account for
damage on the basis of accumulated plastic work implicitly incorporate the effect of strain
hardening on the damage induced in the material (strain hardening materials will exhibit more
plastic work at similar plastic strain ranges or less plastic work at similar total strain ranges).
Because the accumulated plastic work at failure for most materials has been shown to be
variable depending on the loading conditions [12], this class of models may not accurately
predict the accumulated damage.
-
a,
$ 0.0160 I
m
0
-I 0.0150
0
0 a %
0.0140 O O
A 00 ' A
w"
a V
.- 0.0130
sf'
Gi
.- Torsionflorsion
0
c A AxiaVTorsion
* 0.0120
CrJ
a Torsion/Axial ...
.- > 7
s 0.0000
lz
0.6 0.8 1 .o 1.2 1.4 1.6 1.8
Sum of Life Fractions (n,/N, + n.JN2)
Fig. 7. Equivalent plastic strain range in the second load level at the approximate mid-life
point vs. the sum of life fractions

There is a very good chance that competing mechanisms, one linked to the cyclic
deformation behavior and the other linked to crack growth, which may be modeled well
individually but not in a combined sense, are the root of the discrepancies observed in the
literature. Energy and plastic strain based methods do better in modeling crack 'initiation' and